Gender Differences in Campaign Strategies: Lessons from the Nordic Countries

Transcription

1 Gender Differences in Campaign Strategies: Lessons from the Nordic Countries Elin Bjarnegård Uppsala University Hilde Coffé Victoria University of Wellington Pär Zetterberg Uppsala University Paper prepared for the European Conference on Gender and Politics (ECPG) in Uppsala, Sweden, June 11-13, Preliminary draft. Please do not cite or circulate without the permission of the authors. 1

2 Abstract This paper explores whether there is such a thing as a typical male and a typical female political campaign strategy and, if so, what they entail and how the possible gender differences in campaigning can be explained. Our study focuses on four Nordic countries (Finland, Sweden, Norway and Denmark) and draws on the Comparative Candidates Survey (CCS). Our multivariate Logit and Ordinary Least Squares (OLS) regression analyses reveal that male political candidates are significantly more likely to canvas and to give interviews in local and national media compared with female candidates. The gender differences in the presence in the media are especially strong among right-wing candidates. Right-wing male candidates are also substantially more likely to campaign locally compared with their female counterparts. Finally, our analyses reveal that, while campaigning, male candidates focus substantially more on the economy compared to female candidates. We nuance this finding by showing that this is especially the case in parties that can easily be placed on a single left-right dimension. Female candidates are, in their turn, more likely to emphasize non-economic issues, including communication with the constituency and providing services and help to people with practical problems. The paper contributes to the existing literature on gender and representation by exploring the difference of men s campaign issues and strategies in a gender-progressive context. 2

3 Introduction Political campaigns play a major role in political life by influencing electoral support for political candidates and thus electoral outcomes (Hillygus and Jackman 2003). Hence, the gender dynamics within political campaigning, and the subsequent levels of male and female parliamentary representation merit the attention of the academic community. Some studies (e.g., Dabelko & Herrnson 1997; Hernson, Lay and Stokes 2005; Larson 2001) have shown that candidates gender relates to the issues they address during their campaigns. For example, female candidates are more likely than male candidates to campaign on so-called women s issues such as poverty, healthcare, education, child-welfare and family issues (Larson 2001; Dabelko and Herrnson 1997). Most of this research, however, relies on U.S. data and typically scrutinizes the difference of female politicians to the male norm. Furthermore, little, if anything, is known about the extent to which male and female candidates use different strategies during their campaigns. This paper aims to fill this gap and explores gendered campaign strategies in a gender equal political setting: the Nordic countries. In these countries, a fairly gender equal parliament in terms of the number of women is the norm, rather than an unattainable ideal (ranging between 39.1 percent (Denmark) and 44.7 percent (Sweden)). Moreover, women have been present in Nordic politics for a long time. Hence, the Nordic allow us to test whether there is a discernible difference between how male and female candidates run their campaigns, even in a seemingly gender equal context. In testing the extent to which women and men campaign differently in Nordic countries, we look both at the issue focus while campaigning (including the economy and providing services and help to people with practical problems) and at various 3

4 ways of campaigning (such as canvassing, interviews in media, visiting local events, appointments with companies, and debates with competing candidates. To answer our research questions, we draw on the Comparative Candidates Survey (CCS) and focus on almost 3,000 candidates in the four Nordic countries Sweden, Finland, Norway and Denmark. Before turning to a description of our data, we first introduce some relevant theories. Gender and Campaign Behavior Within the field of gender and political representation, the issue of gender differences in legislator behavior has been extensively researched (see e.g. Childs 2004; Franceschet et al. 2012; Swers 2001; Thomas 1991). Research from a various countries has shown that female Members of Parliament (MPs) prioritize different issues and use their representative role differently than their male colleagues. For instance, women in the legislature tend to speak more on feminine issues than men (e.g. Bäck et al. 2014); they author more bills related to gender equality issues than men (e.g. Franceschet and Piscopo 2008; Kerevel and Atkeson 2013); and they focus more on constituency work than their male counterparts (e.g. Thomas 1991). As a consequence of these differences, research has shown that women have commonly been able to have an impact on policy agendas and, to a somewhat smaller extent, also on policy adoption. As more women have been represented in Parliament, a broader range of issues have been subject to political decision making and new laws have been enacted. 4

5 Significantly less attention has, however, been put on potential gender differences among candidates during election campaigns. Yet, analyzing such differences is campaigning is important. First, campaigns influence electoral support for political candidates and thus electoral outcomes (Hillygus and Jackman 2003), which in turn affect the composition of Parliament and thus the levels of female parliamentary representation. Second, during campaigns mandates are created: it is a stage where politicians have the opportunity of choosing how to present themselves, how to communicate with and relate to voters, and which issues to emphasize (Arbour 2013). Empirical research on gendered campaign behavior has mostly focused on the issues candidates address while campaigning. More specifically, research commonly shows that male candidates tend to give more priority to economic issues than female candidates, whereas women, in their turn, pay more attention to social issues (e.g. health care, family issues, etc.) than men (e.g. Kahn and Gordon 1997; Larson 2001; Windett 2014). These differences have been mainly explained using theories of gender issue ownership. Politicians respond to the fact that voters use gender to assess a candidate s policy positions and potential performance, and voters are likely to perceive female candidates as more competent than male candidates on issues such as compassion issues (e.g. income redistribution, service provision) and men as more competent than women on, for instance, economic issues and force and violence issues such as defense (Herrnson et al. 2003). Moreover, gender socialization and life experience may make women more interested in typically feminine issues and men more likely to be interested in typically masculine issues (see e.g. Heath et al. 2005). 5

6 Less is known about the extent to which male and female candidates use different communication strategies during their campaigns. Theories of contextual politics suggest that negative campaigning violates traditional female stereotypes and that female candidates should be less likely than men to attack opponents on most issues (Herrnson and Lucas 2006). Other research, focusing on candidates media participation, has identified generally small differences in male and female candidates use of websites (Carlsson 2007). Finally, looking at the way candidates are presented in television advertisements, the similarities tend to outweigh the differences between male and female candidates (Sapiro et al. 2011; Panagopoulos 2004). In the analysis below, we examine both issue priorities and communication strategies. Based on previous research, we would expect to find substantial differences in issue priorities, but perhaps less so for communication strategies. However, we move beyond a focus on the US that has dominated previous research (see also Lee 2007) and analyze campaign behavior in four of the five Nordic countries (Finland, Sweden, Denmark, and Norway). Analyzing campaign behavior in the context of a relatively gender equal representation such as the Nordic countries is important as it improves our understanding of whether an increased gender equality in representation cements typical male and female behavior among political candidates, or whether women s descriptive and substantive representation instead institutionalizes new issues and new behavior to the extent that male and female politicians behave fairly similarly (and pursue both traditionally male and female issues) during campaigns (see also Lee 2007). In other words, analyzing gender and campaign behavior in a context of 6

7 relatively gender equal representation, our expectations could go in different directions. Our study also adds to previous research by investigating both the campaigning of male and female candidates. Indeed, most previous research on gender and campaign behavior focused exclusively on female candidates and their policy issues and character traits. Although men have been part of empirical analyses, emphasis has been on the way in which women differ from men. However, in order to investigate gender differences in campaign behavior, it is important to acknowledge that men too are gendered beings. Thus, rather than mainly treating men as the invisible norm (Collinson and Hearn 2001), it is important to investigate more generally to what extent women and men differ from one another. This is especially important given the growing levels of female representation. As a consequence, a relevant question is to what extent male candidates still differ from women candidates, also when gender inequalities in legislative representation are (almost) broken. Finally, we build on those who call for an analysis in which other identities are analyzed in conjunction with gender differences (Dolan 2005; Lee 2007). The perhaps most important identity for campaign behavior is party identity (see e.g. Dolan 2005). Campaign strategies and agendas are (to a varying extent) filtered by political parties in virtually any representative democracy: candidates represent a specific party s politics, program, strategies, and goals. As women and men sometimes self-select into specific parties, and thus are more likely to represent certain political parties than others, an observed difference between men and women may in fact be the result of differences in issue preferences and communication strategies between political 7

8 parties. To disentangle the effects, we examine the impact of gender on campaign behavior in the light of the candidate party (see also Dolan 2005). In addition, we test for the possibility that campaign behavior is gendered in some party contexts but not in others by paying attention to the potential interaction between sex and party. For instance, as left-wing parties are generally likely to have a higher proportion of female MPs and be more progressive in gender equality issues than right-wing parties, campaigning might be more gendered in right-wing parties than in left-wing parties. To rephrase, we test the proposition that right-wing male and female candidates differ more strongly in their ways of campaigning than their left-wing counterparts. Data and Measurements To answer our research questions and explore whether men campaign differently than women, we rely on the Comparative Candidate Survey (CSS). 1 The CSS is a crossnational collaboration of standardized surveys with a fairly strong focus on campaign issues, which makes it suitable for our purposes, and possible to identify researchable questions. Over twenty Western countries participate in the CSS in some way, but the data are not comparable across all countries and questions. Therefore, a careful case selection within the CSS has to be made depending on which questions one is interested in. Our study includes four Nordic countries: Denmark (2011), Finland (2011), Norway (2009) and Sweden (2010). The total sample size is 4,038 respondents before deleting 1 See the project s website for more information: 8

9 missing information on the independent variables included in our analyses (see below). After deleting respondents with missing information, the sample size is 3,032 respondents, allowing us to run multivariate analyses with between 2,814 and 2,946 respondents (depending on the dependent variable). There are at least two reasons that make it relevant to focus on these four Nordic countries. First, they all have a long history of relatively gender equal parliaments. For instance, already in the early 1990s, there were more than 30 percent women in each of these countries parliaments. As a consequence, the Nordic countries early on occupied the top four positions (in a global comparison) in terms of women s political representation (IPU 1997). Thus, by focusing on the Nordic countries, we are able to test the proposition that long-term gender equality in parliament leads to a low level of gendered campaign behavior. In that regard, the analysis can be seen as a critical and conservative test of the suggestion about gendered campaign behavior. Second, the countries are similar in both cultural (e.g. Protestantism) and institutional ways, for instance with respect to electoral system (proportional representation with somewhat open lists) and party system (along mostly a single left-right dimension). These shared features make it possible to roughly control for a number of contextual factors already at the outset. Nevertheless, we do include country dummies in our models to filter out the effects of any country-specific factor. For instance, the incentives for politicians to cultivate a personal vote somewhat vary across the four countries (c.f. Carey and Shugart 1995): Finland stands out as having the most open lists. Such institutional features may very well have an impact on the extent to which there is intra-party variation in campaign behavior. 9

10 Dependent Variables Below, we focus both on the ways in which candidates campaign and on the policies they focus on while campaigning. Four measures of ways of campaigning are distinguished: local, media and debate, canvassing, and Public Relations. Our measure of local campaigning includes five campaign activities (Cronbach s Alpha=.70): (1) meeting local party members, (2) visiting local events in business, sports and culture (3) appointments with companies in constituency, (4) appointments with associations and clubs in constituency, (5) organizing and joining large rallies in the constituency. Our measure of media and debate campaigning relies on four items (Cronbach s Alpha=.68): (1) local and regional newspaper interviews, (2) local and regional radio and TV interviews, (3) national radio and TV interviews, and (4) debating with competing candidates in public. 2 Our measure of canvassing relies on one item referring to door-knocking. All items measuring canvassing, local and media campaigning were measured through a six-point scale: (1) no time, (2) 1-5 hours, (3) 5-10 hours, (4) hours, (5) hours, and (6) more than 20 hours. 3 They refer to the hours per week candidates spent personally on the various campaign activities during the final month of the campaign. Since our main interest lies in whether candidates do an activity or not, rather than the time they spent on particular activities, and to keep similarity with the other items measuring the Public Relations way of campaigning (see below) we recoded all items in dichotomous variables, measuring whether the candidate had (1) done the activity or (0) not. Our measure of Public Relations campaigning combines four ways of campaigning (Cronbach s 2 The survey does include a question about the importance of national newspaper interviews, but unfortunately this question was not asked in the Danish sample and could thus not be included in our analyses. Results are however similar. 3 The scale was however different in some countries. In Sweden, there was no category 5 and category 4 referred to hours. In Norway, category 4 referred to hours and category 5 to more than 20 hours. 10

11 Alpha=.82): (1) personal campaign posters, (2) personal ads in the local press, (3) personal flyers, and (4) personal website. The individual items were measured by asking political candidates whether (1) they had used those means or (0) not. Two measures of policy items candidates focus on when campaigning are distinguished: one measuring an economic issue, the other measuring what we refer to as non-economic issues. To measure the focus on an economic issue we use one item asking candidates to what extent they emphasized taking care of the economic wellbeing of the constituency in their campaign. The scale measuring the emphasis on non-economic issues uses three items asking (Cronbach s Alpha=.73): (1) providing services and help to people with practical problems in the constituency, (2) advocating the policy demands of the voters in the constituency, and (3) openness to the voters in the constituency and communicating with them extensively. Answering categories for all items ranged between (1) very much and (5) not at all, but have been recoded in such a way that a higher value refers to greater emphasis. Independent Variables The main focus of the analysis, gender, is a dichotomous variable with the value 0 for female and 1 for male. Further, education is measured as a dichotomous variable indicating whether the individual has attained a university degree. Age is represented by three categories: younger than 46, between 46 and 55, and older than We also include a variable indicating the size of town where the candidate is living. Three categories are distinguished: rural, town, and (suburb of a) city. The latter category is 4 Those three categories were the original categories in the Swedish survey. The year of birth, asked in all other countries, has been recoded to these three categories. 11

12 the reference category in the analyses below. In addition, we add a variable measuring whether the candidate (1) lives in the constituency or (0) not. 5 Besides these socio-economic background characteristics, we also include various political background characteristics. A first political background characteristic indicates the candidate s party affiliation. The parties were categorized in five groups: center-left, center-right, nationalist/populist, green and other. A detailed overview of the different parties included in each category is provided in Table A the Appendix. The center-left is the reference group in the analysis below. We also include a variable measuring whether the candidate was (1) candidate in the previous election or (0) not. Furthermore, we include a measure indicating the candidate s evaluation of his/her chance to win the mandate. The original variable has been recoded into three categories: (1) candidate thought (s)he could not or hardly win, (2) candidate thought it was an open race, and (3) candidate thought (s)he could not or hardly lose. The latter category is the reference category in the analyses below. A final variable referring to the candidate s political background measures whether (s)he (1) had a personal campaign team or (0) not. We also control for the countries as a broad measure of country-level differences. Denmark will be the country of reference in the analyses below. Table 1 provides descriptive information for all variables included in our analyses broken down by gender. 5 We also explored the effect of a variable measuring whether the candidate lived in the constituency or not. As this did not have a major impact and did not influence our overall findings, it is not included in the analyses below. Furthermore, we tested the impact of marital status and employment status. Since these characteristics only had a minor effect on campaigning and did not influence the major conclusions of our study, they are not included in the final analyses presented below. 12

13 [Insert Table 1 About Here] As can be seen in Table 1, men are substantially more likely than women to give media interviews and to canvas. They are, however, equally likely to campaign locally by visiting local events, organizations and business, and to set up a PR campaign by developing personal campaign posters, flyers and websites. Looking at the issues candidates emphasize during their campaigns, this descriptive analysis reveals that male candidates focus considerably more on economic issues than female candidates do. Women, in their turn, are more likely to emphasize non-economic issues, including providing services and help to people with practical problems in the constituency, and advocating the policy demands of the voters in the constituency. Results Having described bivariate links between gender and campaigning, we now turn to a more critical test through multivariate Binary Logistic Regression analyses (for canvassing, which is a dichotomous variable) and Ordinary Least Squares analyses (for all other variables, which are continuous variables). We start by looking at the different ways in which male and female candidates campaign. As stressed in the theory section, there is little previous research that guide our expectations regarding this issue. Table 2 presents multivariate analyses for the four different ways of campaigning explored in the current study: local campaigning, media interviews, canvassing, and public relations. [Insert Table 2 About Here] 13

14 There are no statistically significant gender differences in the extent to which public relations using personal campaign posters, personal ads or flyers or building a personal website are used in the campaign. Reflecting the results found in the descriptive analyses, gender differences in the extent to which male and female candidates give interviews in the media and go from door-to-door are robust and significant. Both these campaign strategies are used significantly more by male candidates. It is worth noting that while media appearance might partly be an effect of gender bias in the media sector male candidates may be invited for interviews more often than female candidates (c.f. Niven and Zilber 2001) the decision to canvas from door-to-door lies with the candidate him/herself to a much greater extent. While no significant (p=.12) gender difference was found for local campaigning in the bivariate analysis presented in Table 1, a minor (p<.10) positive effect of being a male candidate occurs once socio-economic and political characteristics of the candidates are controlled for. In other words, once those characteristics are controlled for, men are slightly more likely to be involved in local campaign activities, such as meeting local party members and visiting local businesses, companies and associations. Additional analyses (not reported here) revealed that this is mainly driven by having a personal campaign team which positively relates to campaigning locally. In other words, if women and men were as likely to have a campaign team, men would be slightly more likely to campaign locally than women. We next turn to the issues that candidates focus on while campaigning. Table 3 introduces two analyses, one on economic issues and the other on non-economic issues such as practical help to constituents and advocating the policy demands of the voters. Earlier research on less gender equal political contexts (ADD REF) has led us 14

15 to expect that male candidates will focus more on economic issues than female candidates. [Insert Table 3 About Here] Table 3 tells us that even in a fairly gender equal political context such as the Nordic countries, male candidates tend to focus significantly more on economy than female candidates. This economic focus seems to be at the expense of other, non-economic, issues, as male candidates tend to focus significantly less on non-economic campaign issues than female candidates do. In other words, this analysis seems to suggest that rather than changing the agendas of male politicians, gender equality in politics has at least to some extent cemented the different issues that men and women emphasize in their campaigns. Having described the main models, we now move on to investigating gender interactions with party affiliation. These models allow us to explore to what extent gender differences in campaigning are similar among the different party families. [Insert Table 4 About Here] Table 4 shows that the interaction between male and center-right party is positive and significant for local campaigning and for campaigning in the media. This means that there is a gender gap in these campaign activities particularly in center-right parties: men in these parties use local campaign strategies and media strategies significantly more than their female party colleagues. Thus, when it comes to local campaigning, 15

16 the interaction analysis reveals gender differences that were barely visible in Table 2. The reason is that the differences are restricted to a set of political parties (i.e. centerright parties), whereas in other parties no such gender gap exist. A closer look at the gender gap in local campaigning within center-right parties reveals that it is mainly caused by center-right women s relatively low levels of engagement in local campaigning (relatively men and women in other parties) rather than center-right men s exceptionally high levels. The same qualification can be made about the findings concerning gendered media use: They are primarily a result of female centerright candidates give fewer interviews than both male center-right candidates and candidates from other parties (male or female). 6 Table 5 present analyses exploring interactions between gender and party family affiliation for the emphasis on economic and non-economic issues. [Insert Table 5 About Here] The introduction of interaction variables in Table 5 nuances somewhat the gendered results relating to issues priorities (as presented in Table 3 above). Perhaps most importantly, the relationship between male candidates and a focus on economic issues does not operate equally in all types of political parties. In green parties, for instance, there is no significant gender difference. In these parties, male and female candidates are as likely or unlikely - to focus on economic issues. A similar pattern is found 6 There is also a significant interaction between being a male candidate and being affiliated with an other party for canvassing. This finding, however, is quite difficult to interpret theoretically, due to the different characteristics of the parties included in the other party group. 16

17 also within other parties as well as within populist parties, albeit at a lower level of significance (p<.1). Thus, the direct relationship in Table 3 appears to be driven mainly by the types of parties that were built around and thus can easily be placed on a single left-right dimension. It is mainly in these parties that male candidates focus substantially more on economic issues than female candidates. Table 5 also reveals that there is a minor negative interaction effect (p<.1) of having a green affiliation and being male when looking at non-economic issues. More specifically, not only do green parties emphasize these issues to a lesser extent than other parties (b=-0.36); among their candidates male candidates focus less on them than their female party colleagues (b=-0.24). Thus, the negative relationship between being a male candidate and focusing on non-economic issues seems to be particularly large among green parties. To illustrate the interaction effects, Table 6 introduces the margins for the main significant interaction effects. [Insert Table 6 About Here] As can be seen from Table 6, the difference between center-right male and female candidates related to local and media campaigning is larger than the difference between male and female center-left candidates. The results also show that centerright female candidates score lower on local and media campaigning than their centerleft counterparts. The interaction models presented in Table 5 revealed that the effect of being male on the likelihood to focus on economic issues was significantly more 17

18 negative among the green candidates than among center-left candidates. The margins confirm that the gender difference in focusing on economic issues is larger among green candidates than among center-left candidates. They further reveal that female green candidates are overall more likely to focus on economic issues than their male counterparts. While higher than among their male counterparts, green female candidates focus on economic issues is still lower than among both male and female center-left candidates. Conclusion The aim of our study was to explore to what extent male and female political candidates campaign differently, both in the issues that they focus on and in the ways they campaign. In line with previous studies (see e.g. Dabelko and Herrnson 1997; Larson 2001; Windett 2014), our analyses focusing on four Nordic countries (Denmark, Sweden, Norway and Finland) revealed that men give more attention to economic issues, even in a context with relatively gender equal political institutions and culture. Interestingly, this finding is not valid for all types of parties, but mainly for those parties that were built around and can easily be placed on a single leftright dimension (center-left and center-right parties, respectively). Furthermore, men were found to put less emphasis on non-economic issues such as advocating policy demands from constituents and providing services and help to people with practical problems in the constituency, than women do. This finding was particularly valid for green parties. Hence, there are topics that men focus less on, even when these issues have been prominent in the political sphere for decades. In other words, men seem to emphasize on the economy at the expense of other issues. They do not focus on more issues but rather on different issues. 18

19 Next to a different focus in their campaign, our multivariate analyses also revealed that male and female political candidates campaign in different ways. Men seem to canvass more and to be more likely to appear in the media and engage more in debates with their political challengers. Moreover, male candidates greater likelihood to be present in the media and in public debates is not common among all parties. Analyses including interaction effects indeed revealed that the likelihood of men being more likely to engage with the media and in public debates is substantially stronger among center-right parties than among other parties. Center-right male candidates are also more likely to campaign locally than their female counterparts, whereas such gender difference does not occur among other parties. By systematically investigating to what extent and how male and female candidates campaign differently, both related to the topics that they focus on and their communication strategies, we believe that our study add some interesting insights to the literature on gender and representation. In particular, it shows that even in the context of relatively egalitarian countries gendered campaigning remains prevalent, both in the issues candidates focus on and in the way they candidate. As always, however, various questions remain unanswered. For instance, we make no comparisons to other contexts and it is therefore difficult to determine whether the differences are substantial as compared to other contexts. Are gendered differences less pronounced in political campaigns in the Nordic countries than elsewhere? This, we do not know. Also, we do not know whether male candidates greater likelihood to be appear in media and engage in public debates has to do with a preferred 19

20 communication style among male candidates or with a media bias that gives male candidates greater opportunity to give interviews and engage in public debates. We also know very little about the way in which male politicians themselves look at their role as men and in relation to female politicians. Analyses of the powerful and privileged are often done from a distance (see Gilding 2010; Madrid 2013) and more qualitative work is needed. 20

21 Table 1 Means/Proportions for All Variables (Standard Deviations in Parentheses) Range a Mean/ Prop. Men St. Dev. Mean/ Prop. Women St. Dev. Sign b Dependent Var. Local (.31).64 (.30) NS 2,814 Media (.31).46 (.31) ** 2,845 Canvassing 0/ % 27.65% ** 2,946 Public Relations (.39).43 (.40) NS 2,905 Economic Issues (1.20) 3.07 (1.22) ** 2,901 Non-Econ. Issues (.96) 3.47 (.87) *** 2,871 Independent Var. University 0/ % 64.20% *** 3,032 Age (ref.: older than 55) Younger than 46 0/ % 42.96% * 3, / % 27.14% NS 3,032 Size of Town (ref.: (Suburb of) City) Rural 0/ % 39.81% * 3,032 Town 0/ % 31.24% NS 3,032 Party Affiliation (ref.: Center-left) Center-right 0/ % 45.95% NS 3,032 Populist/Nationalist 0/ % 8.10% * 3,032 Green 0/1 7.04% 8.10% NS 3,032 Other 0/1 5.57% 3.70% NS 3,032 Candidate Previous Elections 0/ % 17.15% *** 3,032 Likelihood to Win (ref.: Not or Hardly Loose) Not or Hardly Win 0/ % 78.05% NS 3,032 Open Race 0/ % 13.53% NS 3,032 Personal Campaign Team 0/ % 26.59% NS 3,032 Source: Comparative Candidate Survey, Denmark (2011), Finland (2011), Norway (2009) and Sweden (2010) a 0/1 indicates dichotomous variable, 0-1 indicates continuous variable. b Gender difference, Significance tests conducted through OLS and Logit regression analyses controlling for countries. NS: Not Significant; ~ p<.10; * p<.05; ** p<.01; *** p<.001. N 21

22 Table 2 OLS Regression Analyses for Ways of Campaigning Binary Logistic Regression Analyses for Canvassing Local Media Canvassing Public Relations Coeff. St. Error Coeff. St. Error Coeff. St. Error Coeff. St. Error Male.02 ~ *** ** University * Age (ref.: older than 55) Younger than * * ** * * Size of Town (ref.: (Suburb of) City) Rural ** Town.04 ** ~ ~.01 Party Affiliation (ref.: Center-left) Center-right -.04 ** *** *** ***.01 Populist/Nationalist -.07 *** *** ***.02 Green -.08 *** *** ** Other -.14 *** ** ~ **.03 Candidate Previous Elections ** ~.01 Likelihood to Win (ref.: Not or Hardly Loose) Not or Hardly Win -.13 *** *** ***.02 Open Race ** ~.02 Personal Campaign Team.16 *** *** *** ***.02 Country (ref.: Denmark) Finland -.08 ** *** *** ***.02 Sweden.18 *** *** ***.02 Norway.10 *** *** ***.02 Constant.64 *** *** ***.03 Adjusted/Pseudo R-Squared N 2,814 2,845 2,946 2,905 Source: Comparative Candidate Survey, Denmark (2011), Finland (2011), Norway (2009) and Sweden (2010) ~ p<.10; * p<.05; ** p<.01; *** p<

23 Table 3 OLS Regression Analyses for Issue Focus During Campaign Economic Issues Non-Economic Issues Coeff. St. Error Coeff. St. error Male.13 ** ***.03 University ***.03 Age (ref.: older than 55) Younger than ** *** *.04 Size of Town (ref.: (Suburb of) City) Rural.25 *** *.04 Town.14 ** Party Affiliation (ref.: Center-left) Center-right **.04 Populist/Nationalist *.06 Green -.52 *** ***.07 Other -.35 ** Candidate Previous Elections Likelihood to Win (ref.: Not or Hardly Loose) Not or Hardly Win -.32 *** ***.06 Open Race Personal Campaign Team.20 ** ***.05 Country (ref.: Denmark) Finland.17 ~ *.07 Sweden -.33 ** Norway **.07 Constant 3.27 *** ***.11 Adjusted/Pseudo R-Squared N 2,901 2,871 Source: Comparative Candidate Survey, Denmark (2011), Finland (2011), Norway (2009) and Sweden (2010) ~ p<.10; * p<.05; ** p<.01; *** p<

24 Table 4 Interaction Models, OLS Regression Analyses for Ways of Campaigning Binary Logistic Regression Analyses for Canvassing Local Media Canvassing Public Relations Coeff. St. Error Coeff. St. Error Coeff. St. Error Coeff. St. Error Male Party Affiliation (ref.: Center-left) Center-right -.08 *** *** *** ***.02 Populist/Nationalist -.07 * *** **.03 Green -.09 ** ** Other -.11 * * ~.05 Male*Center-right.08 ** ** Male*Populist/Nationalist Male*Green Male*Other * Constant.66 *** *** ***.04 Adjusted/Pseudo R-Squared N 2,814 2,845 2,946 2,905 Source: Comparative Candidate Survey, Denmark (2011), Finland (2011), Norway (2009) and Sweden (2010) ~ p<.10; * p<.05; ** p<.01; *** p<.001. All Models Control for Age, Education, Size of Town, Candidate Last Election, Chance of Winning, Personal Campaign Team, and Country. 24

25 Table 5 Interaction Models, OLS Regression Analyses for Issue Focus During Campaign Economic Issues Non-Economic Issues Coeff. St. Error Coeff. St. Error Male.18 * *.06 Party Affiliation (ref.: Center Left) Center Right *.06 Populist/Nationalist.27 * Green -.28 * ***.10 Other Male*Center-right Male*Populist/Nationalist -.29 ~ Male*Green -.43 * ~.13 Male*Other -.39 ~ Constant 3.24 *** ***.11 Adjusted/Pseudo R-Squared N 2,901 2,871 Source: Comparative Candidate Survey, Denmark (2011), Finland (2011), Norway (2009) and Sweden (2010) ~ p<.10; * p<.05; ** p<.01; *** p<.001. All Models Control for Age, Education, Size of Town, Candidate Last Election, Chance of Winning, Personal Campaign Team, and Country. 25

26 Table 6 Margins for the Main Significant Interaction Effects (Confidence Intervals Between Brackets) Local Media Economic Issues Male Center-left Female Center-left Male Center-right Female Center-right Male Green 2.56 Female Green 2.83 Source: Comparative Candidate Survey, Denmark (2011), Finland (2011), Norway (2009) and Sweden (2010) 26

27 Appendix Table A Categorization Political Parties Party Category Finland Sweden Denmark Norway Center-right National Coalition Party Center Party Christian Democrats Center Party Center Party of Finland Liberal Party Liberal Alliance Christian Democratic Party Christian Democrats in Finland Christian Democrats Liberals Venstre, Liberal Party Swedish Peoples Party in Finland Conservatives Conservative People s Party Høyre, Conservative Party Center-left Social Democratic Party Social Democrats Red/Green Alliance Socialist Left Party Left Alliance The Left Party Socialist People s Party Labour Party Communist Party of Finland Social Democrats Workers Party Social Liberals Communist Workers Party Populist/Nationalist Finns Party Sweden Democrats Danish People s Party Progress Party Change 2011 Green Green League Green Party Other Pirate Party of Finland Feminist Party Liberty Party Future of Finland Independence Seniors Party For the Poor Other Party 27

28 References Arbour, Brian K "Candidate Reputations and Issue Agendas." American Politics Research 41 (6): Bäck, Hanna, Marc Debus, and Jochen Müller "Who Takes the Parliamentary Floor? The Role of Gender in Speech-making in the Swedish Riksdag." Political Research Quarterly 67 (3): Carey, John M, and Matthew Soberg Shugart "Incentives to Cultivate a Personal Vote: a Rank Ordering of Electoral Formulas." Electoral Studies 14 (4): Childs, Sarah New Labour's Women MPs: Women Representing Women: Routledge. Collinson, David L, and Jeff Hearn "Naming men as men: Implications for work, organization, and management." In The Sociology of Masculinity, ed. F. Barrett and S. Whitehead. Cambridge: Polity Press. Dabelko, Kirsten La Cour, and Paul S Herrnson "Women's and Men's Campaigns for the U.S. House of Representatives." Political Research Quarterly 50 (1): Dolan, Kathleen "Do Women Candidates Play to Gender Stereotypes? Do Men Candidates Play to Women? Candidate Sex and Issues Priorities on Campaign Websites." Political Research Quarterly 58 (1): Franceschet, Susan, Mona Lena Krook, and Jennifer M Piscopo, eds The Impact of Gender Quotas New York: Oxford University Press. Franceschet, Susan, and Jennifer M. Piscopo "Gender Quotas and Women's Substantive Representation: Lessons from Argentina." Politics & Gender 4 (3): Gilding, Michael "Motives of the Rich and Powerful in Doing Interviews with Social Scientists." International Sociology 25: Heath, Roseanna Michelle, Leslie A. Schwindt-Bayer, and Michelle M. Taylor-Robinson "Women on the Sidelines: Women's Representation on Committees in Latin American Legislatures." American Journal of Political Science 49 (2): Herrnson, Paul S, J Celeste Lay, and Atiya Kai Stokes "Women Running "as Women": Candidate Gender, Campaign Issues, and Voter-Targeting Strategies." Journal of Politics 65 (1): Herrnson, Paul S, and Jennifer C. Lucas "The Fairer Sex? Gender and Negative Campaigning in U.S. Elections." American Politics Research 34 (1): IPU. Women in National Parliaments: Situation as of 1 January. Accessed on March 15, [cited. Kahn, Kim Fridkin, and Ann Gordon "How Women Campaign for the U.S. Senate: Substance and Strategy." In Women, Media, and Politics, ed. P. Norris. New York: Oxford University Press. Kerevel, Yann P, and Lonna Rae Atkeson "Explaining the Marginalization of Women in Legislative Institutions." Journal of Politics 75 (4): Larson, Stephanie Greco ""Running as Women"? A Comparison of Female and Male Pennsylvania Assembly Candidates' Campaign Brochures." Women & Politics 22 (2): Lee, Francis L F "Gender Differences in Campaign Materials and Politicians' Professional Orientations: The Case of a Hong Kong Election." Sex Roles 57 (11-12):

29 Madrid, Sebastián "Getting into the Lives of Ruling-Class Men: Conceptual Problems, Methodological Solutions." In Men, masculinities and methodologies, ed. B. Pini and B. Pease. Houndmills, Basingstoke: Palgrave Macmillan. Niven, David, and Jeremy Zilber "Hod Does She Have Time for Kids and Congress? Views on Gender and Media Coverage from House Offices." Women & Politics 23 (1-2): Swers, Michelle The Difference Women Make: The Policy Impact of Women in Congress. London: University of Chicago Press. Thomas, Sue "The Impact of Women on State Legislative Policies." Journal of Politics 53 (4): Windett, Jason Harold "Gendered Campaign Strategies in U.S. Elections." American Politics Research 42 (4):